Detail publikace

QUALCOMM-ICSI-OGI Features for ASR

Originální název

QUALCOMM-ICSI-OGI Features for ASR

Anglický název

QUALCOMM-ICSI-OGI Features for ASR

Jazyk

en

Originální abstrakt

Our feature extraction module for the Aurora task is based on a combination of a conventional noise supression technique (Wiener filtering) with our temporal processing technigues (linear discriminant RASTA filtering and nonlinear TempoRAl Pattern (TRAP) classifier). We observe better than 58% relative error improvement on the prescribed Aurora Digit Task, a performance level that is somewhat better than the new ETSI Advanced Feature standard. Furthermore, to test generalization of our approach to an independent test set not available during development, we evaluate performance on American English SpeechDatCar digits and show 10.54% relative improvement over the new ETSI standard.

Anglický abstrakt

Our feature extraction module for the Aurora task is based on a combination of a conventional noise supression technique (Wiener filtering) with our temporal processing technigues (linear discriminant RASTA filtering and nonlinear TempoRAl Pattern (TRAP) classifier). We observe better than 58% relative error improvement on the prescribed Aurora Digit Task, a performance level that is somewhat better than the new ETSI Advanced Feature standard. Furthermore, to test generalization of our approach to an independent test set not available during development, we evaluate performance on American English SpeechDatCar digits and show 10.54% relative improvement over the new ETSI standard.

BibTex


@inproceedings{BUT10471,
  author="Lukáš {Burget} and Stephane {Dupont} and Harinath {Garudadri} and František {Grézl} and Hynek {Heřmanský} and Pratibha {Jain} and Sachin {Kajarekar} and Nelson {Morgan}",
  title="QUALCOMM-ICSI-OGI Features for ASR",
  annote="Our feature extraction module for the Aurora task is based on a
combination of a conventional noise supression technique (Wiener
filtering) with our temporal processing technigues (linear discriminant
RASTA filtering and nonlinear TempoRAl Pattern (TRAP) classifier). We
observe better than 58% relative error improvement on the prescribed
Aurora Digit Task, a performance level that is somewhat better than the
new ETSI Advanced Feature standard. Furthermore, to test generalization
of our approach to an independent test set not available during
development, we evaluate performance on American English SpeechDatCar
digits and show 10.54% relative improvement over the new ETSI standard.",
  address="International Speech Communication Association",
  booktitle="Proc. 7th International Conference on Spoken Language Processing",
  chapter="10471",
  institution="International Speech Communication Association",
  year="2002",
  month="september",
  pages="4",
  publisher="International Speech Communication Association",
  type="conference paper"
}